Americans are an unusually peripatetic people, as both historical and contemporary observers have noted. Before the arrival of permanent European settlers at the start of the seventeenth century, the indigenous population roamed the North American continent after having arrived from Asia more than ten thousand years before (Thornton 2000, p. 9).¹ Soon after the arrival of Europeans, migration was of such concern that permission was required for travel among the early colonies; the ease with which obligations could be escaped through migration led to capital penalties for runaway indentured servants (Smith 1947, pp. 265–70). Two centuries later, Alexis de Tocqueville described Americans' comparative rootlessness: "Millions of men advance at once toward the same point on the horizon: their language, their religion, their mores differ, their goal is common. They were told that fortune is to be found somewhere toward the west, and they go off in haste to meet it” (2000, pp. 268–9). In the twentieth century, Americans were twice as likely to relocate during their lives as the British or Japanese (Greenwood 1997).

High rates of geographic mobility have had important consequences for American economic development. This mobility has facilitated the exploitation of natural resources at locations distant from the narrow band of initial settlement on the Atlantic coast. Farmers moved to more productive land in the Ohio River valley in the late eighteenth century and to the Great Plains by the middle of the nineteenth century. Migrants in the West and the Northwest worked mineral and timber resources. High rates of mobility also spurred the rapid integration of labor markets as transportation improvements – initially canals and riverboats and later railroads – lowered the cost of migration. By the Civil War, much of the gap in wages between the West and the East in the Northern states had been erased (Margo 2000, p. 154). The possibility of migration to cheaper western lands may have also overturned long-standing family support patterns, as children sought their fortunes far removed from parents who were forced to invest in resources other than their children to support themselves in their old age (Ransom and Sutch 1986). And migration from farms to towns and cities in the late nineteenth and early twentieth centuries provided much of the labor force for America's mills, factories, and offices. The migration out of central cities toward suburban counties made possible by streetcars, commuter railroads, and highways has spawned new communities that in many places now compete with nearby central cities for economic supremacy.

Migration's impact can be seen well beyond these economic effects, however. The pressure of population growth in older regions and the need for new territory to which migration could be directed have caused conflict with the indigenous population since colonial times and conflict with other colonial powers (France, Britain, and Spain) over territorial expansion. Since the time of Frederick Jackson Turner, historians have debated the role of easy migration to the west in forging peculiarly American institutions. Though the importance of the frontier as a "safety valve” relieving pressure on eastern labor markets has been debated for more than a century, the role of migration generally, and of the occupational mobility with which it was often accompanied, in preventing the radicalization of the American labor movement has been remarked upon since the time of Marx. High rates of mobility have also influenced rates of civic participation and voting patterns: out-migration has left political power in the hands of a small number of persistent residents, while in-migration has tipped the balance in close elections (Winkle 1988). The shift in population from the Northeast and Midwest to the South and West in the second half of the twentieth century fundamentally realigned American national politics.

Back To Top

Conceptually, the measurement of migration is straightforward.² Ideally, data would identify where individuals were located at some date t, and where they were located at some subsequent date t + n, making it possible to calculate how many had changed locations in the interval (t, t + n) and which places had lost and received population. For example, if the population of locations A and B at time t and time t + n are known, along with how many people moved from A to B and from B to A in the interval (t, t + n), we can easily calculate the gross rates of in-migration and out-migration for each location. Complication arises in the definition of a "location,” the interval between dates, and the populations at the two dates.

The choice of what sorts of places among which migration will be measured will have a large influence on measured rates of migration. For example, if a great many individuals move very short distances (e.g., within the same county), but we are able to examine only moves that cross state boundaries, we will substantially underestimate migration. As long as some moves are over only short distances, the choice of progressively larger geographic units into which to partition the population will result in greater and greater understatement of the extent of migration.³

Similarly, the choice of the time interval (t, t + n) over which to measure migration will influence how much of the actual migration will be captured in measured migration. This difficulty can occur for two reasons. The first is return migration: Suppose that among the population of individuals who move from location A to location B in the interval (t, t + n), some fraction return to location A before time t + n. If individuals' locations are not observed continuously but are observed only at discrete dates (such as t and time t + n), undercounts of both migration from location A and migration from location B will result. Use of smaller and smaller increments n between the initial and terminal dates can reduce this problem. The other difficulty in choosing the interval is that it may be defined imprecisely. For example, it may be possible only to compare each individual's location at date of birth time t with their location at some later date t + n_i , where n_i is the individual's age and differs across individuals in the population. If we compare the rate of "lifetime” mobility calculated in this way across two populations, we may get very misleading results if the age structures of the populations differ substantially: the younger population will have had fewer years over which to migrate than the older population.

Finally, the population for which migration information is available may not be representative of the general population, so the migration behavior observed may not reflect overall migration accurately. Also, information on the population may be inadequate to calculate separately how many have moved in and how many have moved out of each location even if the geographic units and time intervals are not a problem. For example, if it is not possible to identify where the population of a geographic unit was located at a previous date, but it is known that the geographic unit's population has increased since a previous date, net migration can be calculated as the difference between observed population growth and the natural increase of the population (the excess of birth over deaths). This will not reveal precisely how many people moved out and how many moved in; it will show only how the numbers moving in and out compare.

Back To Top

For the colonial period, the calculation of migration rates is limited by the available data. Only a handful of sources provide information on the location of individuals at more than one point in time. For example, colonial militia muster rolls from the French and Indian War and the Revolutionary War describe the place of birth and of enlistment for men of age for military service (Villaflor and Sokoloff 1982). However, this population may be unrepresentative of the entire colonial population, so inferences based on its experiences must be made with caution. Genealogies are another source of information on migrants, though their coverage, too, is narrow (Adams and Kasakoff 1985). For the period through 1790, rates of net migration can be calculated by comparing each colony's population with the predicted excess of births over deaths; the difference measures net total (domestic and international) migration.

For 1800–1860, similar techniques have been used to calculate net migration, but for this period it is possible to separate domestic and international migration (McClelland and Zeckhauser 1982). Applications for land promised in exchange for service in the War of 1812 provide information on the state of enlistment and the state of residence later in life for men who served in the war (Oberly 1986). However, the best information on nineteenth-century migration comes from the Seventh Census of the United States in 1850, which asked for the first time that each respondent give his or her state of birth. Using this information together with the information on the respondent's state of residence, it was possible to examine patterns of internal migration for the U.S. population. "Lifetime migration” (i.e., the migration undertaken between the individual's birth and the date at which the individual is observed in the census) could now be calculated (see Table Ac1–42 and Table Ac43–52). By examining the change in the native-born population in each state together with information on births and deaths (from either separate counts of vital events or the data by age in the census itself), net migration could now be estimated as well (see Table Ac53–205 and Table Ac206–413). Until the end of the Second World War, this was the only source of information available on internal migration for the entire nation. The introduction, in 1947, of the Current Population Survey (a joint project of the U.S. Bureau of the Census and the U.S. Bureau of Labor Statistics) provided additional information: in most years, respondents were asked whether they had moved to a different house, county, or state over a specified time interval (see Table Ac419–427 and Table Ac428–436).

The greatest shortcoming generally in the nation's statistics on internal migration is the lack of pre-1935 data on migration over specified time intervals. Thus, for example, we do not know how many people made intercounty or interstate moves over the course of a year or five years or ten years for any time before 1935. The only information available for the nineteenth century is how many people had moved by the time of a given census out of the state in which they were born. Now we can use new microlevel data from most of the federal censuses from 1850 onward, which report each individual's age, place of birth, and place of residence, and get better results. It is possible to create what demographers refer to as synthetic cohorts and follow their internal migration. For example, we could use the 1850 Census to note the distribution of people born in New England between 1820 and 1830 (these people would show up in the 1850 microlevel data with ages between 20 and 30). We could then look for the same birth cohort in 1860 (when they would be between 30 and 40 years of age) and in 1870 (when they would be between 40 and 50 years of age). In this way, it is possible to say how the distribution of locations for these individuals who all started their lives in New England changed at regular ten-year intervals.⁴

Several studies centered on particular communities have attempted to identify out-migrants by comparing census manuscripts or census-like enumerations (city directories, tax lists, or voter lists) at different dates.⁵ These studies, however, cannot distinguish out-migrations from deaths, as they do not observe individuals who do not appear in the second enumeration. Three studies have linked individuals across successive census manuscripts and made it possible for us to examine migration patterns over ten-year periods that explicitly distinguish the experiences of out-migrants from those of decedents (Schaefer 1985; Steckel 1988; Ferrie 1996).

Back To Top

For the colonial period, only a few broad generalizations are possible. By the end of the seventeenth century, New England was already sending more of its population to other places than it was receiving people from other colonies and from abroad; from 1680 to 1780, the region saw a net loss of 52,000 people (Gemery 2000, p. 171). It appears that geographic mobility rates increased over the last quarter of the eighteenth century. Between 1771 and 1798, the fraction of men in a sample of Massachusetts genealogies who had not moved from their place of birth fell from one half to one third, as the fraction moving 100 miles or more rose from one eighth to one third (Adams and Kasakoff 1985, p. 367). In militia muster rolls from the French and Indian and Revolutionary Wars, migration rates also seem to have increased over time: in the four colonies for which muster rolls were examined in both wars (New York, Pennsylvania, Maryland, and Virginia), the fraction of native-born recruits who enlisted in the county where they were born fell from the French and Indian War (1754–1763) to the Revolutionary War (1776–1783). Mobility was generally higher in the South, and a substantial urban-to-rural migration among the native-born can be seen (Villaflor and Sokoloff 1982, pp. 542, 562).

Between 1800 and 1860, migration rates appeared to increase further. The predominant trend was movement from east to west, with the Middle Atlantic states (New Jersey, and the eastern portions of New York and Pennsylvania) providing most of the migrants to the Northwest. East–west migration was greater in volume in the Northern states than in the Southern states, and there is evidence of some south–north migration in the decades prior to the Civil War (McClelland and Zeckhauser 1982, pp. 6–7).⁶

Despite their shortcomings, the data on internal migration over the century and a half since the first questions on place of birth were included in the 1850 Census of Population document several important general trends in internal migration. Figure Ac-A shows the fraction of the native-born population residing in the state of birth at each census from 1850 to 1990. The first regularity is the clear rise since 1900 in the fraction of the population located outside the state, division, and region where they were born. This fraction rises with only one interruption – in the 1930s, when it falls slightly. Before 1900, there is a decline, suggesting that before the era of increasing mobility in the post-1900 period, there was a decrease in mobility from the eve of the Civil War to 1900.

This measure of lifetime migration is somewhat misleading, however, if the age structure of the population changes over time. And the United States was clearly a population with more young people in the second half of the nineteenth century than in the second half of the twentieth. If attention is restricted to individuals of a particular age at each census, a somewhat different picture emerges. Among white native-born males of age 55 at the time of the census, the fraction residing outside the state of birth was 45 percent in 1850 (those born in 1795). By 1900 (for those born in 1845), this fraction had fallen to 40 percent, and it reached its nadir in 1940 (among those born in 1885) at 30 percent. By 1990, it had climbed to 39 percent, still substantially below its level in the 1850–1880 period. Lifetime interstate migration thus seems to be somewhat less common at the end of the twentieth century than in the second half of the nineteenth century (Hall and Ruggles 1999, Figure 3a; see also Rosenbloom and Sundstrom 2000, Figure 4).

Table Ac-B reports net migration by region for three eras (1850–1900, 1900–1940, and 1940–1990) for the native-born white and black populations. These were calculated using the "census survivor” technique.⁷ The most striking patterns here are: (1) the persistent population losses among whites suffered by the New England and Middle Atlantic states, and even by the East North Central states (the old Northwest) from 1850 onward; (2) the large flows of white population into both the Pacific and South Atlantic states since 1940; and (3) the onset of the "Great Migration” out of the South seen in the data for blacks, with out-migration doubling in volume for the South Atlantic states and tripling in volume for the East South Central states between 1850–1900 and 1900–1940, and then nearly tripling again for the East South Central states and rising by a factor of six for the West South Central states between 1900–1940 and 1940–1990.

Since 1920, the U.S. Department of Agriculture has followed migration from farms to nonfarm locations (see Table Ac414–418). The rate of net migration away from farms in Figure Ac-C shows a great deal of year-to-year fluctuation that reflects economic conditions in agriculture. In only three years – 1932, 1933, and 1946 – is the rate negative. The former reflects back-to-the-farm migration in the depths of the Depression, whereas the latter reflects the return of World War II veterans and those employed in wartime industries to the farm. The average rate of migration away from farms rises sharply from 2 to 3 percent per year before World War II to roughly 5 percent per year after the war.⁸

The Current Population Survey (from 1947 onward) and the Census of Population (from 1940 onward) provide more detailed information on short-distance migration. The percentages changing county since 1935 show remarkably little trend over time (see Figure Ac-D). The only substantial difference is between the 1940 figures for migration over five years (which report pre–World War II migration between 1935 and 1940) and the entire postwar period. If anything, the American population seems to have become slightly less mobile since the mid-1980s.

A large number of studies have been completed for specific communities in the United States that give us a sense of geographic mobility in the nineteenth century that is similar conceptually to that contained in the Current Population Survey and the censuses from 1940 onward. The ten-year nonpersistence rates (the fractions of individuals not located in the same county over a decade) for mid-nineteenth-century communities were 56 percent in eleven cities (1850–1880) and 64 percent in nine rural counties (1850–1880). Higher rates were observed for lower-class workers before World War I, but higher rates were then observed for white-collar workers after World War I (these studies are surveyed in Thernstrom 1973). These studies are unable to distinguish out-migration from deaths, however, and tell us nothing about where out-migrants have gone.

An alternative approach is to create data like that in the contemporary Current Population Survey by linking individuals across successive censuses. In one study, this has been done for 4,938 native-born males linked at 1850–1860 (see Ferrie 1996). Of the nearly five thousand males located in both 1850 and 1860, 47 percent changed counties between these dates (Ferrie 1996, Table 5).⁹ Unfortunately, none of the intervals in the Current Population Survey series are as long as ten years. However, there are intervals of one and five years reported, as shown in Figure Ac-D. Based on these rates, over a ten-year interval in the post–World War II period, roughly 37 percent of those present in the United States in both years would have changed county.¹⁰ Another comparison is possible: among males age 20–29 in the 1850–1860 linked sample, 56 percent changed county over a decade; in the young men cohort of the National Longitudinal Survey, the rate of intercounty mobility among males age 20–29 was only 49 percent over the period 1971–1981. This suggests that intercounty migration rates may have been more than 25 percent higher in the 1850s than they were from the 1950s onward.

The availability of linked data for the nineteenth century makes possible a final comparison with more recent data: rates of gross migration among regions. Table Ac-E shows interregional flows for the 1850–1860 decade and interregional flows in the five years preceding the 1970, 1980, and 1990 Censuses. In both the 1850s and the last third of the twentieth century, New England is a region of low population turnover: even though it lost one tenth of its native-born adult male population in the 1850s and one fifth of it over a typical five-year period between 1965 and 1990, it experienced very little in-migration. The East North Central and West North Central states experienced high turnover in the 1850s (for example, with one quarter departing the East North Central states, but large inflows from the Middle Atlantic states and the West North Central states). By the twentieth century, these regions experienced much lower turnover. In the 1850s, in both the North and the South, rates of out-migration from east to west increased; by the twentieth century, this pattern was no longer apparent (e.g., New England lost more population over five years than the Middle Atlantic or West North Central states; the South Atlantic region lost more than any region except the Mountain and Pacific regions). Finally, gross flows reveal a somewhat surprising flow from south to north in the decade before the Civil War: a substantial fraction of those who began the 1850s in the East South Central and West South Central regions moved by 1860 into the East North Central and West North Central states. In the late twentieth century, this flow reversed direction: population moved from the Middle Atlantic to the South Atlantic, from the East North Central to the East South Central, and from the West North Central to the West South Central, though this latter flow was smaller than in the 1850s.

Back To Top

1850–1860: series Ac19, Ac21, Ac26, and Ac28; thereafter, series Ac5, Ac7, Ac12, and Ac14.

Note that the data cover only the white population for 1850–1860. See Table Ap-G in Appendix 2 regarding the composition of census regions and divisions.

Back To Top

[TABLE Ac-B Net migration of the native-born population, by region, race, and major time period: 1850–1990]

Back To Top

The negative of series Ac416 expressed as a percentage of series Ac414.

Back To Top

Series Ac419, Ac424, Ac428, and Ac433.

The years on the horizontal axis represent the beginning of given time intervals.

Back To Top

[TABLE Ac-E Gross migration of the native-born population, by region: 1850–1860 and the late twentieth century]

Back To Top

Adams, John, and Alice Kasakoff. 1985. "Wealth and Migration in Massachusetts and Maine: 1771–1798.” Journal of Economic History 45 (June): 363–8.

Ferrie, Joseph P. 1996. "A New Sample of Americans Linked from the 1850 Public Use Micro Sample of the Federal Census of Population to the 1860 Federal Census Manuscript Schedules.” Historical Methods 29 (Fall): 141–56.

Gemery, Henry A. 2000. "The White Population of the Colonial United States.”  In Michael Haines and Richard Steckel, editors. A Population History of North America. Cambridge University Press.

Greenwood, Michael J. 1997. "Internal Migration in Developed Countries.”  In M. R. Rosenzweig and O. Stark, editors. Handbook of Population and Family Economics. Elsevier.

Hall, Patricia Kelly, and Steven Ruggles. 1999. "Moving through Time: Internal Migration Patterns of Americans, 1850–1990.”  Unpublished (November).

Margo, Robert A. 2000. Wages and Labor Markets in the United States, 1820–1860. University of Chicago Press.

McClelland, Peter D., and Richard J. Zeckhauser. 1982. Demographic Dimensions of the New Republic: American Interregional Migration, Vital Statistics, and Manumissions, 1800–1860. Cambridge University Press.

Oberly, James W. 1986. "Westward Who? Estimates of Native White Interstate Migration after the War of 1812.” Journal of Economic History 46 (June): 431–40.

Ransom, Roger, and Richard Sutch. 1986. "Did Rising Out-Migration Cause Fertility to Decline in Antebellum New England? A Life-Cycle Perspective on Old-Age Security Motives, Child Default, and Farm-Family Fertility.”  Working Papers on the History of Saving number 5 (April).

Rosenbloom, Joshua, and William Sundstrom. 2000. "Long-Run Patterns of Interstate Migration in the United States: Evidence from the IPUMS, 1850–1990.”  Unpublished (November).

Schaefer, Donald. 1985. "A Statistical Profile of Frontier and New South Migration: 1850–1860.” Agricultural History 59 (October): 563–78.

Shryock, Henry, Jacob S. Siegel, and associates. 1980. The Methods and Materials of Demography. U.S. Government Printing Office.

Smith, Abbott Emerson. 1947. Colonists in Bondage: White Servitude and Convict Labor in America, 1607–1776. University of North Carolina Press.

Steckel, Richard. 1988. "Census Matching and Migration: A Research Strategy.” Historical Methods 21 (Spring): 52–60.

Thernstrom, Stephen. 1973. The Other Bostonians. Harvard University Press.

Thornton, Russell. 2000. "Population History of Native North Americans.”  In Michael Haines and Richard Steckel, editors. A Population History of North America. Cambridge University Press.

Tocqueville, Alexis de. 2000. Democracy in America. Translated by Harvey C. Mansfield and Delba Winthrop. Chicago: University of Chicago Press.

Villaflor, Georgia C., and Kenneth L. Sokoloff. 1982. "Migration in Colonial America: Evidence from the Militia Muster Rolls.” Social Science History 6 (Fall): 539–70.

Winkle, Kenneth J. 1988. The Politics of Community: Migration and Politics in Antebellum Ohio. Cambridge University Press.

Back To Top